My dissertation has two goals: the first is to prove that historical syntax has a value in subgrouping; the second is to reconstruct parts of the syntax of the Aramaic dialect groups. The work focuses on three topics: the development of adverbial subordination, nominal modifiers (relative clauses and demonstratives) and speech marking. The approach used is historical-comparative. Since historical syntax has never been applied systematically to any Semitic languages, not to mention to the family as a whole, and since it has been argued to be without value in reconstruction and subgrouping, the work opens with a long methodological explanation and discussion of mechanisms of change, which will be shown to operate in Semitic. The first chapter surveys types of subordination in Semitic, North-West-Semitic (the direct ancestor of Aramaic) and Aramaic. It is argued that while Old Aramaic did not deviate from North-West-Semitic, later dialects differ significantly from early Semitic as well as closely related Semitic languages, such as the Canaanite languages and Arabic. It is shown that Aramaic developed a rich system of subordination, mainly on the basis of prepositions, while other languages used nouns. It is also shown that the Aramaic relative particle is an obligatory non-matrix marker, which is not the case in other Semitic languages.

The second chapter surveys the position of the demonstrative pronoun in Semitic and Aramaic, and explains the deviations found in the syntax of Aramaic. The chapter further look at the phenomenon of resumption. It is argued that the pattern arose independently in Aramaic and spread in a predictable manner from the genitive construction to the preposition and from there to the verbal object. The syntax of the relative clause is explained as a type of noun modifier. The third chapter deals with the syntax of speech marker, especially direct speech. It is shown that the developments described in chapter 2 apply here too; the relative pronoun marks direct speech as a part of its extended function as a non-matrix marker. It is also argued that Aramaic does not attest to a quotative particle.